Variability-aware RRAM PDK for design based research on FPGA/neuro computing

Lead Research Organisation: Liverpool John Moores University
Department Name: School of Engineering

Abstract

The semiconductor industry has provided the devices we have enjoyed for many years, including mobile phones, personal computers, on-line banking etc. The growing functionality of these products is a result of making the components, namely transistors and memory elements, ever smaller, at the rate that in every 18 months or so the number of components in a given area has doubled, which also makes the devices run faster. The industry now runs into a fundamental roadblock in shrinking the devices further, so we need to look for a new device type which will continue to provide higher performance. One strong contender is the RRAM (resistive random access memory) which we will investigate in this project. This device can be programmed to offer either a high or low electrical resistance: that is, store a logic "0" or "1", or even with some intermediate levels in between. It can store information which will remain even after the power is turned off, as so called non-volatile.

With this device, a number of disruptive developments are under intensive research world-wide. Its first potential application is to increase the speed of the non-volatile memory chip in computers by more than 10 times and provide potential for further increase in the number of components. The second is in the artificial intelligence (AI) computing which mimics the functionality of human brains. AI has been widely used by Google, Facebook, Apple, etc. RRAM has the potential to bring a breakthrough in AI by solving the density, connectivity and memory bandwidth limitations of AI hardware based on conventional devices. The third is to revolutionise the programmable computing with its smaller size and non-volatility, providing advantages for computing in data centres and Internet of Things, in which the vast amount of data will be streamed through internet and the scalability and energy efficiency provided by RRAM become critical.

The behaviour of RRAM devices, however, is stochastic, meaning that a large variation occurs during the device operation. At present, the lack of systematic understanding of the variability and the missing tools for variability-aware simulation hinder the research progress in RRAM-based circuit and systems design for neuromorphic and programmable computing. In this project we will collaborate with UK's leading IC design company, ARM Holdings, and the world no.1 EDA software company, Synopsys, providing direct insight into the fundamental properties of RRAM variability and developing a predictive variability-aware product design kit (PDK) that can be directly used within commercial EDA software by designers, enabling the research and design of novel RRAM based neuromorphic and programmable computing systems. We expect this project to have a significant direct impact on the UK and global ICT industry in the forthcoming Artificial Intelligence (AI) and Internet of Things (IoT) era.

Planned Impact

This project aims to address the critical simulation and characterisation of variability issues in resistive memory devices (RRAM), which place fundamental limits on the performance of its application in neuromorphic and programmable computing. The global IC market amounts to more than $335 billion in 2015. IC design plays an important part in the UK Electronic Systems Community that employs more than 850,000 people nationally in 30,000 enterprises, constituting 2.9% of the workforce, a vital enabler to wide market domains such as aerospace, defence, healthcare, retail, media and education. In this project we will collaborate with the leading company in the IC design markets, ARM and Synopsys, and we expect direct impact on the UK and global IC industry. Novel RRAM variability-aware product design kit will be developed in this project and disseminated into our industrial partners and other leading institute such as imec. Circuit/system design industry will benefit from this project by using the PDK as an enabler for innovative circuit design in neuromorphic and programmable computing. The predictive variability-aware model developed in this project should benefit circuit designs wherever RRAM is used, which allows systematic exploration of novel circuit topologies through simulation with commercial EDA tools. Detailed knowledge of device level variability issues in advanced manufacturing processes is now recognised as a key requirement for fab-less companies including those in the UK. The combination of theoretical modelling and characterisation used in this project will provide a test-proven approach in selection and evaluation of material, process and device structure, and provide useful guidance for the RRAM community in general. The research has the potential to put the UK at the forefront of a crucial emerging area in the coming AI and IoT era, which is likely to impact on the global economy as well as on the UK and European economy. The project has excellent potential to generate wealth in the UK by the worldwide sale of simulation tools and design IPs to the leading companies.
 
Description RRAMs are strong candidates for new generations memory technology. Filament rupture/restoration induced by movement of defects, e.g. oxygen ions/vacancies (Vo), is considered the switching mechanism in HfO2 RRAM. However, details of filament alteration variability during switching are still speculative, due to lack of experiment-based probing techniques to directly monitor its spatial and energy profiles and to correlate them with the switching/failure mechanism, impeding its understanding and modeling.

In this project, for the first time, an Random-Telegraph -Noise (RTN) -based defect tracking technique (RDT) is used to monitor the defect movements and the spatial and energy (XT, ET) profile of the critical filament region (CFR). CFR alterations can therefore be directly correlated with switching variability, and new relaxation and endurance failure mechanism in OTS has been revealed. and the endurance in GeSe OTS has been improved by more than 5 orders. This technique has been taken up by industrial partners at world-leading research centre imec, providing a useful tool for RRAM technology development.

For the first time, the following key advances have been achieved: Varibility in both filamentary and non-filamentaty RRAM devices have been characterised and used to model its impact on neuronetwork accuracy. Switching and relaxation Mechanism in GeSe OTS devices have been investigated with both experimental and theoritical evidence, and its endurance has been improved by more than 5 orders a new program scheme . True Random Number Generator and Stochastics Computing have been implemented using the GeSe OTS devices.
Exploitation Route This techniques have been taken up by industrial partners including Intel, Samsung, and Micron at world-leading research centre imec, providing a useful tool for RRAM technology development. A paper has been presented at the flagship conference IEDM 2016, VLSI 2019, respectively. A number of papers have been published in leading journals.
Sectors Electronics

URL https://www.ljmu.ac.uk/about-us/staff-profiles/faculty-of-engineering-and-technology/department-of-electronics-and-electrical-engineering/wei-zhang
 
Description Resistive-switching devices are based on materials in which the resistance can be varied and memorized by controlling the flow of electric charge in thin dielectric layers. These devices will lead to a range of disruptive inventions and the technology is expected to become one of the major driving forces to overcome the scaling limits at the end of roadmap, supporting the ICT industry for the next 20 years, which has opened multiple revolutionary frontiers for ICT industry. The non-volatile resistive-switching devices and selector devices have high speed (ns), high density (4F2), long retention (10 years), high endurance and 3D integration, therefore will not only drastically enhance the performance of non-volatile memories, but will also revolutionize computer architecture. Its simple structure, low cost and 3D compatibility with back-end of line processing warranty its system-on-chip application which can be embedded within the CPU. It also has a high potential to replace DRAM, Flash memory and HDD all together for its fast speed (DRAM), non-volatile memory (Flash) and high capacity (HDD), as referred to the "Storage Class Memory (SCM)". This offers a host of new opportunities to system designers, opening up the possibilities for ultra-fast operation with truly persistent data, providing a significant increase in information throughput beyond the traditional benefits of scaling. It can also provide nano-sized, programmable multi-level variable resistance for a number of analogue applications including neuromorphic computing system in which its nonlinear dynamics can be used as synapses/neurons, a breakthrough for artificial intelligence computing. Its disruptive impacts on ICT industry are widely predicted. The outcomes of this projects have been used at imec and by its core industrial partners including Intel, Samsung, Micron, etc. to develop the next generation of RRAM technology.
First Year Of Impact 2018
Sector Digital/Communication/Information Technologies (including Software),Electronics
 
Description IMEC memory consortium, Belgium 
Organisation Interuniversity Micro-Electronics Centre
Country Belgium 
Sector Academic/University 
PI Contribution A LJMU PhD student have been trained at IMEC for 3 months each year and participated in the weekly meeting at IMEC. His progress have been closely monitored. This provides a direct channel for collaboration between the present project team and IMEC consortium.
Collaborator Contribution IMEC Industrial Affiliation Program is a world-leading consortium including virtually all of the top semiconductor manufacturers among its core technology partners (eg Intel, Samsung, Micron Technology, Hynix, STMicroelectronics, NXP Semiconductors, and Toshiba, etc.) as well as most of the major capital equipment developers and manufacturers. Partners delegate senior industrial researchers to IMEC to carry out joint research using best in-class equipment within IMEC's 300mm clean room infrastructure to solve the most important issues commonly faced by the industry. The collaborative agreement officially signed by LJMU and IMEC takes the advantage of this program, which allows us to work closely with these partners throughout the project. This provides a unique and most effective path to disseminate the research results directly to the major industrial companies, and ensures the project to work at the front line of research and development of resistive switching devices. For example, 1) The project will be steered by researchers from the industrial consortium at IMEC to ensure its industrial relevance; 2) The 300k Euro worth of samples provided by IMEC are manufactured to the highest standard according to industrial partners' requirements in order to investigate issues of major concerns; 3) A LJMU PhD student will be trained at IMEC for 6 months each year and participate in the weekly meeting at IMEC. His progress will be closely monitored. This will provide a direct channel for collaboration between the present project team and IMEC consortium; 4) The IP-irrelevant results can be disseminated directly to the core industrial partners through these meetings and feedbacks are received directly from the partners. In these meetings the potential applications of new theory and characterisation techniques are discussed and taken up by partners. 5) Senior managerial and technical members of partners participate the review week held twice every year at IMEC, during which the latest development will be presented and disseminated to the senior level, and will have direct impact on the research and development decision making. Four of the world top five semiconductor manufacturers, Intel (1st), Samsung (2nd), Micron (4th) and Hynix (5th), are members of the IMEC consortium and are working together on the resistive switching device programme. The fact that IMEC is willing to provide a wide range of support and become the project partner and Micron and Samsung are willing to endorse the project and become the members of the steering panel demonstrate the importance of the issues addressed in this project to the global industry, and their recognition of the project team's ability to make major breakthrough. On this basis, it is clear that the project will have an international reach and will strengthen the link between UK and world leading companies. The results will provide important support to the modelling, manufacturing, quality assessment, and reliability sections of the industry, and provide the possibility to greatly enhance its technological and scientific impact.
Impact A number of papers have been published in the flagship journal IEEE Transactions on Electron Devices and conferences IEEE Electron Device Meeting 2019 and IEEE VLSI Technology Symposium 2019, and the technique has been taken up by industrial partners at imec. Two research visits to imec has been carried out by LJMU researchers in 2019. A number of online project meeting have been held in 2020.
Start Year 2018
 
Description UK variability consortium 
Organisation Interuniversity Micro-Electronics Centre
Country Belgium 
Sector Academic/University 
PI Contribution The team at LJMU has focused in the past 20 years and in five EPSRC grants on developing new techniques for characterising defects in dielectric stacks, including HfO2 and Al2O3. A number of new techniques have been developed, such as the fast single pulse, multiple-pulse, on-the-fly techniques, etc. Systematic investigation has been carried out and new technique has been developed for the resistive-switching RRAM devices. A number of papers have been published in the flagship journal IEEE Transactions on Electron Devices and conferences IEEE Electron Device Meeting 2019 and IEEE VLSI Technology Symposium 2019, and the technique has been taken up by industrial partners at imec.
Collaborator Contribution The partners have complementary strengths on the modelling, synthesis and characterisation of dielectric thin film: University of Glasgow on theory and simulation of defects in oxides and variability. A combination of modelling and simulation has been used at UoG to identify combinations of materials and the influence of their defect structures on key resistive switching variability parameters. These have been characterised using a range of physic-chemical-electrical analytical methods to correlate their microstructures with the modelling data and the electrical performance of resistive switching devices. IMEC has provided state-of-the-art RRAM devices to LJMU, and new techniques has been developed at LJMU, which has been taken up by industrial partner at IMEC, including Intel, Samsung and Micron.
Impact A number of papers have been published in the flagship journal IEEE Transactions on Electron Devices and conferences IEEE Electron Device Meeting 2019 and IEEE VLSI Technology Symposium 2019, and the technique has been taken up by industrial partners at imec. Two posters and two presentations have been delivered at the IMEC technology review week to the industrial partners. Two visits to imec by LJMU researchers have been carried out in 2019. A number of online project meetings between LJMU/imec and LJMU/Glasgow have been held in 2020.
Start Year 2018
 
Description UK variability consortium 
Organisation University of Glasgow
Country United Kingdom 
Sector Academic/University 
PI Contribution The team at LJMU has focused in the past 20 years and in five EPSRC grants on developing new techniques for characterising defects in dielectric stacks, including HfO2 and Al2O3. A number of new techniques have been developed, such as the fast single pulse, multiple-pulse, on-the-fly techniques, etc. Systematic investigation has been carried out and new technique has been developed for the resistive-switching RRAM devices. A number of papers have been published in the flagship journal IEEE Transactions on Electron Devices and conferences IEEE Electron Device Meeting 2019 and IEEE VLSI Technology Symposium 2019, and the technique has been taken up by industrial partners at imec.
Collaborator Contribution The partners have complementary strengths on the modelling, synthesis and characterisation of dielectric thin film: University of Glasgow on theory and simulation of defects in oxides and variability. A combination of modelling and simulation has been used at UoG to identify combinations of materials and the influence of their defect structures on key resistive switching variability parameters. These have been characterised using a range of physic-chemical-electrical analytical methods to correlate their microstructures with the modelling data and the electrical performance of resistive switching devices. IMEC has provided state-of-the-art RRAM devices to LJMU, and new techniques has been developed at LJMU, which has been taken up by industrial partner at IMEC, including Intel, Samsung and Micron.
Impact A number of papers have been published in the flagship journal IEEE Transactions on Electron Devices and conferences IEEE Electron Device Meeting 2019 and IEEE VLSI Technology Symposium 2019, and the technique has been taken up by industrial partners at imec. Two posters and two presentations have been delivered at the IMEC technology review week to the industrial partners. Two visits to imec by LJMU researchers have been carried out in 2019. A number of online project meetings between LJMU/imec and LJMU/Glasgow have been held in 2020.
Start Year 2018
 
Description Invited Talk 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact IEEE 15th International Conference on Solid-State and Integrated-Circuit Technology (ICSICT), 2020
Year(s) Of Engagement Activity 2020
URL http://www.icsict.com/
 
Description Invited talk 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact Invited talk at an international conference
Year(s) Of Engagement Activity 2019
URL https://www.lboro.ac.uk/media/wwwlboroacuk/external/content/schoolsanddepartments/physics/documents/...
 
Description Workshop 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Invited talk at the Annual Workshop 2019 of 111 Project Base of Wide Band-gap Semiconductor & Micro-nano Electronics at Xidian University from July 15 to 17 2019. 100 postgraduate students attended and had active discussions.
Year(s) Of Engagement Activity 2019